- Chinese models are cheaper and lighter then their equivalent US models
- American firms prefer the lower operating costs and model portability
- DeepSeek is now worth around $60 billion
According to a new South China Morning Post report, an increasing number of US companies are now adopting Chinese models and tools like DeepSeek amid broader cost concerns.
It seems that DeepSeek offers significantly lower costs that leading US models developed by firms like OpenAI and Anthropic, while delivering the performance that they need.
As a result, DeepSeek has climbed to the top of several developer rankings as businesses continue to cut costs and increase efficiency.
DeepSeek is becoming popular because it's so... cheap?
According to the report, the trends reflects a much broader shift in the AI market, with companies prioritizing performance less and emphasizing cost more, all in a bid to strike the best and cheapest balance.
DeepSeek is also a popular option because of its open-source approach – unlike many UK alternatives, companies can download, customize and deploy DeepSeek models on their own infrastructure, which helps to reduce dependency on external providers. The report details how many Silicon Valley startups have recently turned to Chinese models.
Chinese companies have also noticed this, and are now releasing lighter models and charging less than typical US companies.
The company has also attracted attention for building competitive models using fewer resources and more efficient training techniques compared with rivals, which is a good thing given US restrictions on advanced semiconductor exports to China.
DeepSeek also recently secured $7.4 billion in additional funding, with Tencent and CATL contributing considerable amounts each, leaving DeepSeek with a valuation of around $60 billion.
Looking ahead, an upcoming V4 model is expected to offer even better cost efficiency as international customers turn to the Chinese firm for cheaper models.
